Upload
mangilal-saraswat
View
280
Download
0
Embed Size (px)
Citation preview
Enhancement of Palm Print
using Median Filterfor Biometrics Application
(A presentation on latest research made by researchers in biometrics area)
Presenting by
Mangi Lal (14/CC/417) M.Tech. student in High Performance Computing
at Computer Center Departmentof National Institute Of Technology Durgapur,
West Bengal, India
01/05/2023 2
01/05/2023 3
01/05/2023 4
01/05/2023 5
01/05/2023 6
01/05/2023 7
01/05/2023 8
01/05/2023 9
Introduction continues…
• The pre-processing plays a vital role
• Paper presents a new technique to pre-process
• Focuses on palmprint boundaries extraction and enhancing the palmprint ridges
01/05/2023 10
Introduction
• Experiment results shown Enhanced palmprint has bright ridges
• Shall give better verification
01/05/2023 11
•David, Z. et al. proposed online palmprint identification system
•Feature extraction using circular Gabor filter
•Normalized hamming distance used to measure similarity
•Own developed realtime palm print acquisition device
RELATED WORKS
01/05/2023 12
•Samples of 7,752 from 193 peoples in different days of two different size 384 x 284 and 768 x 568.
•The larger size was resized to 384 x 284.
•They did total number of 30,042,876 matching.
•The number of correct matching was 74,086 (98%)
RELATED WORKS
01/05/2023 13
•0.04 false acceptance rate and the corresponding threshold 0.3425
•System equal error rate 0.6
•Total execution time 0.6 seconds
RELATED WORKS
01/05/2023 14
• Haruki Ota et al.’s remote system using touchless palmprint recognition algorithm-1. Extraction of an image, 2. Reduction of the extracted image, 3. Flesh color detection based on HSV color system, 4. Opening, 5. Detection of key points and 6. Extraction of palmprint region
RELATED WORKS
01/05/2023 15
•They collected 60 images from 12 people
•Computed False Rejection Rate (FRR) for all 120 combinations and False Acceptance Rate (FAR) for all 1650 possible combinations
•FAR is higher rate 55.7% and FRR is 1% and the corresponding threshold is 0.182
RELATED WORKS
01/05/2023 16
•H. B. Kekre et al., proposed a palmprint recognition technique using Kekre’s wavelet’s energy entropy based feature vector
•384 different people’s palmprints was available in the database of Hong Kong Polytechnic
•They enrolled 125 persons from this database
RELATED WORKS
01/05/2023 17
•More than 5200 tests for intra class matching were done
•Euclidian distance and relative energy entropy were calculated for each test
•The relative distance lies in the range of 15 to 65 for genuine palms and others lies in the range of 45 to 135
RELATED WORKS
01/05/2023 18
Wei Li et al., proposed an efficient technique for joint 2D and 3D palmprint matching using alignment refinement
Hemantha K. K., et al., proposed a technique to identify palmprint based on wide principal lines
David Z. et al., proposed a technique for multispectral palmprint verification
Hoang Thien Van et al., proposed a palmprint verification using GridPCA for Gabor Features
RELATED WORKS
01/05/2023 19
Proposed Enhancement Technique
01/05/2023 20
Canny edge detection
• multi-stage algorithm to detect a wide range of edges in images– A Gaussian blur is applied to clear any speckles
and free the image of noise– A gradient operator is applied for obtaining the
gradients' intensity and direction– Non-maximum suppression determines if the pixel
is a better candidate for an edge than its neighbours
– Hysteresis thresholding finds where edges begin and end
01/05/2023 21
Median filter
• The median filter is normally used to reduce noise in an image
01/05/2023 22
Experiment
• Database of palmprints of IIT Delhi considered
• 1391 palmprint taken in the format of bitmap
• Original: 800 x 600 px and the auto cropped 150 x 150 px
• categorized into left hand and right hand
01/05/2023 23
01/05/2023 24
Experiment Result
• enhanced edge image was more accurate
• The false rejection rate and false acceptance rate found decreased
• Time average was 0.40 milliseconds
01/05/2023 25
Conclusion
• Simple and faster
• Several applications and devices coming with
• Proposed enhancement algorithm is better than the normal edge detected algorithm
• Planned to develop system based on this in the near future
01/05/2023 26
My review regarding this paper
• Excellent paper• Understandable language• Discussion is direct to point
• The false rejection rate and false acceptance rate analysis not given for proposed method
POSITIVE
NEGETIVE
01/05/2023 27
References of this presentation
• The discussed research paper• Papers available at ieeexplore.ieee.org• wikipedia.org (images and text)• photobucket.com (images)• broadsheet.ie (images)• bp.blogspot.com (images)
01/05/2023 28
The discussed research paper• Published in-
International Conference on Advances in Science and Technology (Special issue: ICAST2014)
• Presented in-International Conference on Advances in Science and Technology (ICAST2014) on 15 Feb 2014
01/05/2023 29
And the paper authored by
Sir Chandran Saravanan
01/05/2023 30
Any query